Tumor clonogen repopulation during the course of fractionated irradiation and proficient DNA repair have been suggested as two major causes of radiotherapy failure for head and neck carcinomas. Consequently, the investigators hypothesize that agents interfering with repopulation (e.g. inhibitors of DNA replication) and repair of radiation damage offer potential for improving therapy providing normal tissue toxicity is not also enhanced. Results of preclinical studies show that fludarabine, an adenine nucleoside analog, inhibits chromosome repair, yields more than additive tumor cytotoxicity when combined with radiation, and has a positive therapeutic index. Therefore, they propose studies to test combination of radiation and fludarabine in patient with locally advanced head and neck cancer and assess pharmacologic and radiobiologic determinants of late normal tissue toxicity of combined therapy. Gemcitabine, a second nucleoside analog which also has been shown to inhibit DNA repair, is potentially attractive in combination with radiotherapy because it has a long triphosphate half-life in tumors and activity against human tumors frequently refractory to radiation alone (pancreas, rectal, lung, and head and neck). Therefore, the last specific aim is to determine the optimal schedule of administration and the therapeutic ratio of the combination of gemcitabine and fractionated radiotherapy in murine tumor and normal tissue model systems. The investigators feel that these proposed studies will test the hypothesis that clinical outcome (local control) can be improved by combining nucleoside analogs and fractionated radiation in order to address the problems of tumor repopulation between fractions and radioresistance due to competent repair process. Thee studies will also begin to identify biomarkers which might help in identifying specific therapeutic strategies for particular mechanisms of radioresistance.